Computational Statistics

HIGHER DIMENSIONS 169

Exercise 4.6

Modify the following function pairslm() so that it calculates the resid-

uals of the regression of all original variables in matrix x by regression

after the new variable x$fit and produces a scatterplot matrix of these

residuals.

pairslm <- function(model, x, ... )

{ x$fit <- lm(model, x)$fit; pairs(x, ...) }

Add title, legends, etc.

Use the “trees” data set as an example.

We discussed the transition from p

0

to p

0

+ 1 variable. The scatterplot matrix allows us to get a

quick overview of a (not too large) number of candidates (three potential additional regressors

in our case). The transition from p to p −1 for the elimination ...